Estimating the Value of a New Postsecondary Credential: An Audit Study of Community College Bachelor’s Degrees

This study will focus on CCB programs in six cities across three fields (CS-IT, Business, and Nursing) and the returns to these degrees in their associated local labor markets. The six local labor markets are spread across three states (Austin, Houston, and Dallas in Texas; Cape Coral and Miami in Florida; and Seattle in Washington) that are the leading awarders of CCBs across the country. Each of the CCB programs chosen has a comparable associate degree offered by the same community college as well as a comparable traditional bachelor’s degree at a nearby non-selective four-year college. Our program choices represent similar programs situated in geographically unique labor markets that allow us to better explore heterogeneity across several axes. Although both Texas, Florida, and Washington each introduced CCB legislation nearly 20 years ago, Washington and Florida have granted universal CCB approval and all community colleges in the state offer bachelor’s degrees. Florida’s CCB growth happened much earlier and more rapidly than Washington’s and Texas lags behind both states in terms of number of community colleges offering BAs and the number of BAs offered at each college conditional on CCB approval. These conditions will allow us to understand the value of a truly “new” credential in the eyes of employers across settings that vary in general familiarity with CCBs. As in our pilot study, we plan to pair the resume audit study with a post-experiment employer survey to further explore employer recognition, understanding of degrees, and credentials.

We will submit resumes to jobs located in the largest metropolitan area in each state near the selected CCB institutions (Austin, Houston, and Dallas in Texas; Cape Coral and Miami in Florida; and Seattle in Washington). Job searches will encompass all areas and neighborhoods within the associated metropolitan statistical area (MSA). We will identify open positions by searching once per day for a period of several weeks for vacancies to identify eligible jobs. We will collect the job title, employer name, location, salary, benefits information (if given), and any other information provided by the posting before submitting a randomly generated set of resumes.

2026-05-26

2026-12-31

Callback rates

We consider "callbacks" to include any form of positive employer follow-up, including email responses. We will track both explicit interview requests and a general callback measure that includes personalized requests for more information as well as interview requests.

Fictitious applications will be sent to two types of jobs: those that require an AA or below, and those that require a BA. When applying to jobs that require an AA or below, we will send three resumes that vary by degree type: an associate’s degree, a CCBA, and a BA from a nearby open-access public four-year institution. When applying to jobs that require a BA, we will send two resumes that vary by degree type: a CCBA, and BA from an open-access four-year. This will allow us to estimate heterogeneity across jobs with different education requirements. We will also vary applicants’ perceived race and ethnicity (as signaled by their name) to estimate how employer discrimination may interact with differential returns to degree type.

Not available

Randomization done in office by a computer

Resume/fictitious applicant

No

14000-18000

14000-18000

Applications will be sent to two types of jobs: those that require an AA or below, and those that require a BA. When applying to jobs that require an AA or below, we will send three resumes that vary by degree type: an associate’s degree, a CCBA, and a BA from a nearby open-access public four-year institution. When applying to jobs that require a BA, we will send two resumes that vary by degree type: a CCBA, and BA from an open-access four-year. The number of total applications submitted will be constrained by the number of available job openings but given our power calculations we aim to send a minimum of 14,000 resumes across all states and fields of study.

We plan to submit applications to 14,000-18,000 job postings. Table 1 in the attached analysis plan shows the required sample sizes for a range of minimum detectable effects sizes (MDES), assuming various baseline callback rates. Sample size calculations were conducted using the J-PAL Power Calculator with a significance level of 0.05 and power of 0.80. Our baseline callback rates come from previous similar audit studies (Darolia et al., 2014 and Deming et al., 2016) and from our own pilot study, where 21% of applicants received an interview request. We expect callback rates to be more in line with past literature (4-8%) than our pilot study, since the pilot focused on a tight labor market (early childhood education), but the full study will target industries similar to past literature. The estimates in Table 1 imply that even with the most conservative callback rate, we will be able to detect precise effects of 0.02 by submitting resumes to 17,400 job postings. The estimates in Table 1 give sample size requirements to detect main effects. We are also interested in heterogeneous effects by field of study, race/ethnicity, gender, and labor market. The most conservative power analyses would multiply the sample sizes in Table 1 by the number of groups for each heterogeneity analysis: for example, assuming a callback rate of 0.08 and aiming for a MDES of 4 percentage points, we would apply to 1,634 job postings to detect main effects, but 1,926*3 = 5,778 job postings to detect heterogeneous effects across our 3 fields of study. Similarly, we would apply to 5,778 postings to detect heterogeneous effects by race/ethnicity, or 17,334 (1,926*3*3) postings to detect effects by race/ethnicity and field of study. Thus, our target number of 14,000-18,000 job postings will allow us to detect precise main effects under very conservative callback rates, as well as moderate effect sizes across multiple layers of heterogeneity under expected callback rates.